Modeling Corporate Credit Climate Risk

By Patrick Naim and Laurent Condamin
Assessing the effects of climate change on the banking sector is becoming increasingly urgent. While public perception of the risk posed by climate change has ebbed and flowed within a fairly fixed range, policymakers in the Biden administration are focused on climate change and in using tools like financial regulation to address climate-related risks in the economy. Moreover, the COVID-19 pandemic has shown bankers the operational risk posed by a “black swan” event that creates unexpected economic turmoil.
First and foremost, climate change increases physical risks. This increase is systemic in nature, and will affect the economy—which will in turn condition anticipatory policy changes. These new policies will affect the economy and create a transition risk. Banks will be primarily affected by the climate and political impacts on the economy, and to a lesser extent by the transition and climate impacts they will experience directly.
All these changes interact with each other, and how the world will look as a result is highly uncertain—especially at more distant time horizons. To make informed strategic risk management decisions, banks would be wise to represent this uncertainty in their models.
We present a method for assessing the impact of climate change on a bank’s corporate credit portfolio built on three ideas:
  1. Corporations are sensitive to physical risk and transition risk, depending on their geographic locations, industry sector, and balance sheet structure.
  2. Climate sensitivity ratings, coupled with one particular climate scenario, allow modelers to calculate a multiplier to be applied to the corporations’ probability of default, or PD.
  3. For a bank, the possible extent of future climate scenarios and the structure of its portfolio and its evolution, will condition its stress on credit risk.

Corporate borrowers’ sensitivity to climate change

The notion of corporate climate sensitivity is straightforward:
  • Physical risk. The increase in natural disasters or extreme climate events could impact the corporate facilities directly, and its entire supply chain. These events have many consequences, including increased direct costs and production delays.
  • Transition risk. A company that is subject to a tax on greenhouse gas emissions incurs an additional cost of transition.
These risks and additional costs will affect companies’ earnings. Not all companies, even those in similar industries, will be equal in this respect.
For transition risk, companies with high greenhouse gas emissions will be hit harder by the carbon tax. Transition sensitivity can be defined as the ratio of the current cost of emissions to earnings. This ratio is highly variable. For example, we observed a ratio of 1 percent for a retail company, 3 percent for an oil company and 5 percent for an airline. If the carbon tax were multiplied by 10, this would correspond to a decrease in earnings of almost 50 percent for the airline, all other things being equal. Therefore, this would impact its financial stability.
For physical risk, the reasoning is similar, as an increase in natural disasters or extreme weather events will impact the company’s profits. But there is an important difference. Once we have made an assumption about the future price of carbon, say $100 per ton, and knowing the structure of the company’s balance sheet, we will know the effect on profits. But, if we assume an increase in natural disasters, say by a factor of two, we only know that a company has an increased risk of lower earnings. Evaluating the impact of physical stress for a company is similar to calculating an operational risk VaR for a bank. This is a risk assessment exercise that can only be performed by the company itself, as it requires detailed knowledge of its locations, key facilities, suppliers and key customers. However, in the absence of this detailed analysis, a physical risk sensitivity indicator can be defined as the ratio between the current frequency of natural disasters and their anticipated frequency in the future. In a way, the model starts by stating that the company is currently adapted to withstand its natural risks, and then the stress is captured by the change in risk level.
Thus, we define the two indicators of the company’s sensitivity to climate risk as follows:
  • The sensitivity to transition is the ratio between emissions and profits.
  • The sensitivity to physical risk is the expected increase in natural disasters in the regions where the company primarily operates.

Calculating the credit risk multipliers

Now we need to transform these sensitivity indicators into default probability multipliers. To do this, we will use the Merton model and the Black-Scholes equations in both cases. Without going into technical details, the Merton model allows us to relate the value of a firm’s assets and debt to its probability of default, and the Black-Scholes equations allow us to deduce the value of a firm’s assets from the value of its equity.
For transition risk, the approach is simple. An increase in the price of carbon will have an impact on a company’s profits, modulated by its sensitivity to the transition. Assuming a stable price-to-earnings ratio, the decline in earnings will result in an equivalent decline in the market value of the company. Using the Black-Scholes equations and the Merton model, we in turn infer the change in the firm’s asset value and its probability of default. Our calculations show that the transition PD multiplier ranges from two to four for carbon intensive industries.
For physical risk, the approach is a trickier. First, we consider that the current level of natural disasters is already priced into the current probability of default.  This is the mathematical way of stating that the company is currently adapted to resist its natural risks. Then, knowing the current probability of natural disasters to which the firm is exposed, we deduce the contribution of natural disasters to the probability of default. Finally, we only need to stress the probability of disasters to evaluate a stressed PD. For physical risks, our calculations show that the stress on PD ranges from 10 percent to 20 percent when the frequency of natural disasters doubles and from 20 percent to 50 percent when the frequency quadruples.

Measuring portfolio climate stress

At this point, we can consider assessing the overall impact of a climate scenario on a bank’s corporate credit portfolio. By now, we have three ratings for each company in the bank’s portfolio:
  • A credit rating
  • A transition sensitivity rating
  • A physical risk sensitivity rating
Even if climate scenarios are very complex to address in their entirety, we can get an idea of how they will be taken into account by the financial system from the first exercises conducted by the Bank of France and the Bank of England. The transition risk is represented by a carbon cost trajectory that could be modulated by region, and the physical risk is represented by an emissions scenario that conditions the temperature. Research from the Intergovernmental Panel on Climate Change links each scenario to an increase in natural disasters by region.
The diagram below shows the interrelationship of all these elements. 
The diagram above is a Bayesian network showing the dependencies between the variables. A bank has a credit analysis for each client and third-party organization noting its PD, its loss given default (or LGD) and exposure at default (or EAD). Thanks to its new climate risk analysis team, it has assigned a physical and transition risk sensitivity rating to each corporate client. These ratings are input in the bank’s climate scenario. This climate scenario defines the cost of carbon and the increase in natural disasters, and it allows the bank to calculate separate PD multipliers for the transition risk and the physical risk.
The Bayesian network can also be used for simulation. The simulation can be run either for a single given scenario, as requested by a regulator, or by keeping the scenarios uncertain, if a bank wants to work on its own risk assessment. The approach presented above allows banks to take a structured and progressive approach to assess climate risk stress on their corporate credit portfolios.
We recently tested this approach in the context of a preparatory regulatory exercise in Europe. This exercise was conducted on a large, diversified portfolio, and we showed that the impact of transition risk— while individually higher for carbon-intensive companies, was dominated by the impact of physical risk at the portfolio level, since a larger fraction of the portfolio was exposed to physical risk. With the information available at the date of the exercise, the impact of the transition risk was approximately 15 percent on the total expected credit loss, while it was between 15 percent and 45 percent for the physical risk. These results are highly dependent on the portfolio structure.
Our quantification method is based on simplifying the assumptions related to the structure of the corporate client’s balance sheet and main geographical locations of its physical assets. However, this quantification can be refined, if more precise information is available—for example, information on the bank’s future projects that could mitigate its transition risk, or its exposure to physical risk.
At a time when regulatory pressure is increasing for banks to be prepared to assess climate risk, we believe that our approach is a rational and scalable to banks’ profiles and needs.
Patrick Naim is CEO, and Laurent Condamin is managing director, at Elseware, which produces a structured risk scenario assessment tool known as MSTAR. The authors are partners in ABA’s Structured Scenario Analysis Benchmarking Working Group.